-
So i thought of starting the cluster by deploying the ingress, which is where traffic will be coming through. allowing http network traffic
naturally there was a conflic inside me between caddy and nginx, i have never liked the long configurations for nginx yet the only little part i care about is the proxy configurations. But yet again i havnt used caddy on kubernetes -
I start working on caddy on kubernetes.
-
Can i do it? 😉. lets see.
-
I find most of the info i need on https://medium.com/@ramontayag/caddy-on-kubernetes-619e21e8fbbd
-
so i come to find out that caddy apparently has a problem where the certs populating accross hosts cause the application to freeze and not accpt any requests. probably not cool. So i get an idea to checkout traefik. i liked the docs i read but i think last time i tried using it,i was probably too dumb to understand.
- First i have to read about the Cluster roles, i think the traefik guys are big on security inside kubernetes and i think its good thing for me to know, one can never know where that knowlegde can save or hurt(my cluster nodes hacked or needing to help a company with a security problem)
- After reading abit and trying to do a traefik deployment, i realize that the kube yyaml files arent always working, ie samples from teh internet. i think its due to version changes and updates. so i decide to try out helm, which is an opensource package manager for adding applications to custers as if they were packages in a js app through
yarnandnpm - Since i already had helm installed in teh system, i realize that i needed to add a repo and init helm on the cluster(i really want to deploy this thing locally for testing), hopefully it doesnt wase my time by not working out. :sad:
- I start to remember very clearly that there is a time when i really wanted to know how to repecate k8's clusters for testing, and here i am realizing that helm charts are the thing that i wanted. i think the entire repo "mono-repo" will be a helm chart. and then i can deploy teh helm chart on a cluster but with a different name on cli deployments(ofcourse running on a fuckin test cluster. because on CI we need to be able to push to a demo cluster and release frequesntly, activate and drop features. so i gues helm is what i need, rather than a service hosted by some other company - teraform(ill be back to you big boy))
- So im reading about ho to make helm charts and how they work, yeey 😄. Im'ma be a pro.
Im starting to think helm is the best idea ever, its relatable to the rancher store where you can upload docker-compose.yamls to do quick deploys of entire applications, only this doesnt have a fancy UI's and forms and buttons. Its just a bunch of downloadable files, not sure if there are provisions for asset files and description metadata to be used as if anyone decides to build an api on top of it? hehe this OSS projects. If they do have that then i gues they are dope.if they dont maybe its something i could work on 🎈
- learns that helm repos are literarry any https server that can service public files. This means that even a git repo that can server raw files can be a chart repo! but i can imagine literary just running hs on any folder now makes it able to serve helm and become a source of truth for CI. I think this is similar to registries, just that registries are for blobs and hence why we store them in S3. files that will be changed by humans are stored on git so that they can auto deploy on changes
- Girlfriend came and lets just say i couldnt do shit for like 21 hrs, sigh. anyway. got to write this on tommorows log lets meet on the 2nd Feb